Is there a way to do sampling on GPU directly? like
x = torch.cuda.LongTensor(100,100).random_(100)
The other way seems slow:
x = torch.LongTensor(100,100).random_(100).cuda()
Is there a way to do sampling on GPU directly? like
x = torch.cuda.LongTensor(100,100).random_(100)
The other way seems slow:
x = torch.LongTensor(100,100).random_(100).cuda()